The use of pilot models in dynamic performance and rotor load

EIGHTEENTH EUHOPEAN HOTOHCHAFT FOHUM

E · 1 5 Paper No.143

THE USE OF PILOT MODELS IN DYNAMIC PERFORMANCE

AND ROTOR LOAD PREDICTION STUDIES

J.C. IIAMM

~ESTLAND HELICOPTERS LIMITED

YEOYIL

SOMERSET, ENGLAND

September 15-18, 1992

AYIGNON, FRANCE

TilE USE OF PI LOT MODELS IN DYNIIM I C PEHFOJ\MANCE liND HOTOH L0/\0 !'HEll I CTION STUDIES

J.C. llamm

Westland Helicopters Lld Ycovil, England

IIBSTRIICT

This paper describes the Westland method of using a helicopter engineering simulation, controlled by a pilot model, for dynamic performance and rotor load prediction studies. The reasons for using a pilot model are explained and current and future uses of the models are given. The aims and philosophy of pilot modelling are presented and the method of usc for performance prediction studies is outlined; including the methods used to

validate the model, and to generate the performance data for inclusion in

the rotorcraft flight manual. The structure of the Westland pilot model method is given and the capability of the method is illustrated by examples.

1. Introduction

Helicopter manufacturers are required to promulgate airfield performance data in the rotorcraft flight manual. The data must be based on flight test experience, but some means of interpolating between the test cases is necessary. To generate the data base for all of the conditions required for certification, Westland use an engineering model of the helicopter, controlled by a simulation of the helicopter pilot. The purpose of the computer model is to accurately predict the flight path which a helicopter would follow, when flown to the flight manual technique, in a given set of circumstances. To achieve this, the pilot simulation has to observe the same vehicle limitations and piloting constraints as the human pilot.

The need for dynamic performance models is not new1 nor is the method described here. The earliest Westland dynamic performance model, of which I am aware, was used to determine the take-off performance of the Wessex 60 Series 1 in the 1960's. It was written in Elliott Autocode for the Elliott 803 computer. From this first model, a suite of two-dimensional (longitudinal symmetrical) flight path simulations evolved. Programs were written, in FOHTRAN, to model the Lynx and Seaking, and numerous manoeuvre specific versions were used during the Westland 30 certification process. By 1985 the code was becoming outdated and

expensive to maintain and work was begun on the creation of an entirely new

and completely general longitudinal flight path simulation. This was written as a modular program which would lend itself to the simulation of new manoeuvres, by engineers not fully familiar with all aspects of the code. The rotorcraft was fully described by the input data sets and the model was able to simulate any conventional helicopter type, flying any longitudinal manoeuvre. Now known as H/\PS the Westland "Helicopter Airfield Performance Simulation'', the program is used for all of Westland's longitudinal flight path prediction ;10rk.

Also in 1985, discussions between Westland Helicopters and the Royal /\erospace Establishment (now Ill\/\ llerospace) idcn ti f icd a rcqui rcrncn t to

study a manoeuvring helicopter rotor. It was decided t!1at a new simulation should be created: to lnvestieate rotor behaviour and performance in manoeuvres, for the prediction of dcsien loads, and lo confirm stability augmentation system features. !'rcviously, rotor loud cases were rtln by

defining a manoeuvre using a simple simulallon model lo c~;t~\bl \sh lhe rotor conditions, and lhcn examining the rotor behavior using a separate analysis program; each condition requiring up to t.wenly minutes of nm time. As the analysis of a complete manoeuvre was a lime consumins and expensive procedure, on 1 y essen t 1 a 1 cases could be cons ide red. The pr~oposcd new analysis program. Hhich is now known as lhe Coupled Hot~..)t' Fu~_:;clage ~1odcl,

or CHFM, would overcome these difficulties by incorpor~1l.ing a manol~uvrc

capability, with significant improvements to Lhc rotor analysis program. By coupling the dynamic systems of tlw rotor and fuseL\~~e. using complex rotor modes. the analysis would accommodate Lhe effects of hub motion on rotor load and vibration prediction. As the intcnti.:.>n was to analyse manoeuvring flight, an algorithm was required to gent;~rate the control inputs to "fly" the simulation through m0.nocuvres. Aft.ct~ reviewing the

possible alternatives, the Westland pilot model method was selected, and

work was commenced in April of 1989 to extend the logic used in tl1e two dimensional models, to the much more complex task of controlling a full

three dimensional simulation.

The resulting Cf{F'M pilot model, which is now

running but has not yet been validated,

is known as the "Helicopter

Manoeuvre Simulation Manager" or HELMSMAN.

I t

has been written as a self

contained module which accepts vehicle response as input and generates

control positions as output.

A more complete description of the Coupled

Rotor Fuselage Model can be found in references 1 to 4

2.

Overview of Pilot Modelling

There are many reasons for using a pilot model. Simulations

controlled by pilot models are inexpensive to run, easy to modify, give

repeatable results and eliminate the variability inherent

in piloted

simulations.

By removing the human element from the control loop, they

obviate the need to run the helicopter simulation in real time; this has

several advantages.

A simulation which can be run at faster than real time

is of great benefit when generating data for the multitude of cases

required for flight manual charts.

On the other hand, the ability to run

at much less than real time is an absolute necessity if you wish to use

affordable computers to run complex models; hence the need for a pilot

model

to control the CRFM for rotor loads predictions in manoeuvring

flight.

Furthermore, pilot models give the user a clear insight into what

is going on.

The engineer has full control over the simulation, can change

any vehicle or handling parameter at will, and can repeat cases as often as

necessary. Because the simulation can be run on a workstation, without the

need for a cockpit, pilot, visuals etc., the ability to study manoeuvring

flight can be made available to the engineer,

at his desk, at very

reasonable cost.

At

Westland,

the Aerodynamics Performance Group use helicopter

engineering simulations, controlled by a pilot model, at almost every stage

of vehicle development.

The suitability of the programs for parametric

studies make them valuable tools at the preliminary design stage; for

example, when sizing the main rotor for acceptable vertical reject

performance.

Once flight testing gets under way, the abilily to try out

handling techniques, and examine the consequences of vehicle limitations.

can be used to forewarn the flight crews of any potential problems.

Areas

of high risk can be analysed in great detail. For this work, lhe per·formance prediction simulation is complementary to the piloted engineering simulation. As the vehicle dcveloprnc~nt cycle continues, and certification testing begins, the use of the dynamic performance prediction models becomes intense. At the beginning of the certification process, the models are used to optimise the piloting lechniques for best performance. Preliminary charts are produced, as a target to aim for during the certification flying - the benefits of this should not be undercslirnatcd. As soon as the handling techniques have been <lpproved, and the tnodeJ. has been validated against the flight t.e~-;t results, the simul;1tion rnay be w;ed

to generate the extensive datac;ets which will be plotted to produce t.he dynamic performance charts in the rotorcrnft flight manual.

The models have also been found to be highly suitable for research work. The flAPS program was recently used by Westland, in a study for the United Kingdom Civil Aviation Authority (CAA), to cxamlnc the engine fai 1 ure performance of he 11 copters operating to offslwrc pla l. forms. An example of a HAPS generated engine failure flight profile is presented in figure 1.

Computer generated pictorial

representaion of HAPS predicted

flight path and fuselage attitude

Figure 1 : HAPS predicted offshore platform flyaway.

Earlier this year, HAPS was used to calculate the engine failure performance of the Lynx Mk. 9 and it is currently being employed to study the airfield performance of the EHlOl in support of the certification program. Both HAPS, and a HELMSMAN controlled blade element model of the EHlOl, are used from time to time for vehicle development studies; for example, to predict control range requirements, blade lag ranges and transient torque requirements. In the future, the Coupled Rotor Fuselage Model, controlled by the HELMSMAN pilot model, will be used to calculate helicopter dynamic performance, to determine stressing cases, and for the

prediction of rotor performance and loads in manoeuvres.

The primary aims in creating pilot simulation models are:

a) To observe all of the vehicle and piloting limitations which would constrain the performance of the vehicle when flown by a human pilot. b) Thereby to generate realistic flight paths.

3. Method of Use of pilot model controlled simulations for performance

prediction studies.

When creating a dynamic performance prediction model, the first action is to create an aircraft input data set for the helicopter and engine combination to be modelled. In the early stages of an aircraft program no flight vehicle will exist, and the most that can be done to validate the model is to compare the steady state power requirements wilh other theoretical predictions. At this stage the simulation will only be used for preliminary design studies or to make initial performance estimates. When flight testing begins, Lhe aircraft input data scl is brought up to date, to incorporate any changes, ond lo model any ~-;pcclal equipment on the test aircraft.

The steady state performance of the model is confinnc'd by means of a power carpet match and, if available, an analysis of llh' powc,- lH·cakdm.Jn. At Westland we aim to achieve as close a malcl1 as possitle to tl1c mcastlrc<i power carpet, using lhe basic power prediction model. Th~._'n, for~ performance prediction work, we use a look up table of correction Clcl.ors lo give an exact rna tch to the measured power carpel. For power ;.\\"ail able \.Je usc a

look up table of installed power, obtained by running the engine manufacturer's deck, \.Jith allowances for .in~;t.allation losst~s.

At the same time, consideration is given to lht"' manoeuvres to be

flown. The technique required to fly each manocuvrQ is ~1nalyscd and complex manoeuvres are broken down into phases. For t.~~1ch stage of the manoeuvre, the piloting and vehicle constraints and the logic switching triggers are identified and the manoeuvre subroutines arQ coded and tested. \Jhen developing code to define a handling technique it is beneficial to

involve a pilot. A workstation with good graphics cap3bility, which can run the simulation and display the vi tal performance ~-..~1rameters in real time, has been found to greatly reduce the time and cost associated with this task. Experience has shown that the run-time displays should present information in an analogue form which can be quickly and easily assimilated by pilots and engineers. A representation of the cockpit instrumentation and either a simple outside world view or a time hist.:.:ry trace has been found to work well (see figure

2).

As there is no requirement to include

Figure 2 : Interactive display. the human pilot in the control

resolution and refresh rate are

loop, the specifications for the display relatively low, which helps to keep the hardware costs down.

spent in optimising studies may also be chosen technique and margins.

Once the basic manoeuvre has been programmed, the handling technique for best performance. made to check the variability characteristics

thereby to establish the size of the required

time i.s Abuse of the safely

11anoeuvre trials are flown to determine lhe f/::rforrnancc of the vehicle and to demonstrate to lhe cerlificulion ~1uttlorilic:s lh~l Lt1e specified handling technique is simple to fly and f.~lves repeatable performance. It is quile likely lhat, for one r·ea!:;on or another, ~-;uch as

visual cueing or airspeed indicator behaviour, the handling tcchn\qu,, will change during the trials and that the final technique, which w\ 11 be described in the flight manual, will differ slightly from the \n1tla1ly defined technique. for this reason it is useful to have the capability to update the model, and re-issue the target performance charts, during lhc trials. The move from mainframe computers to woi~kstat ions \.Jhich c~u1 be transported to the trials site, should make this easier.

With the vehicle steady state performance confirmed by lhe pO\..rer carpet match, and measured dynamic manoeuvre results available; tl1c accuracy of the modelling of the vehicle dynamic response, the fidelity of pi lot model and the validity of the complete package as a pe1·formance prediction tool, can now be verified. The vehicle dynamic response model is validated by comparing the predicted vehicle res;,onse to attitude changes and control inputs, with the measured response of the helicopter on flight test. The method is to run the simulation in a matching mode, so that, instead of the pilot model generating the co:1trol inputs, the fuselage attitude and collective stick positions measured during the flight trials are fed into the program as input. The resulting prediction of the

"

rJ28(\

t: -10 240 '-40 120 0 I

I

• : I I

I I I I I I I I

I I J I

FUS~r._[GE

PITCH AmTUDE

I

I

I I

I I I I I I I

I I I

I

I I ,{

I VI

I

I

I I I I I~ I I I

I /

I

I

I

/

I I

I I

rn

UE AIRsPEED ·~· .. .,. _

r

1 , 1 1 1 1 ' f"' \ I . I I I \

f-+-+-+-+-+'<1--+-+-l ,...-'f:_._,f--1--+-+-+-+--INote: Flight test and simulation -~H---1

f--1--+-+-+-+--P...._...,...,q--+-l--1-f--l-1 +-+--+-+-blade collective pitches

are..L-~

f-1--+-+-+-+-+++-+--+--+--1-f--1--11--+-+-+--referenced to diff~rent radial

r-I

0 4

ENGINE T()R()UE

!

stations I I I I I I

I I

I I

I

I I I

I""'-

I

I I I I

1\

.•. _.,_ MAIN ROTOR BLADE"-1-1

-t+-++-f--1

COLLECTIVE PITCH I ,.I I

I I I I I

--!

I

F(JLED ENGINE TORQUE I I

_{I I}

_I I

I I I I I I

8 12 2 4 2 8 3 2 3 6

TIME (SECS)

40 48 52 56 60

Figure 3 : HAPS matching of a Lynx hover flyaway manoeuvre.

helicopter response is then compared v.1ith the: measured rr::sponsc:s of various key parameters .i.n order to asses the fidelity of thr: dynamic response model. The comparison is achieved by displaying the measured parameters on a computer screen and overlaying the predicted trar::es, as they arc generated by the simulation. figure 3 presents a !lAPS matching of a Lynx hover flyaway manoeuvre.

It should be noted that, when t.hc simulatlon i~;: run in matching mode, there is no feedback of vehicle response to the pilot. model; the simulation is open loop. Any change in the fllghl conditions, such as may

be caused by variations ln wind speed etc., which are Jh'l rec()rdcd and so are not modelled, will cause the predicted and mcasurc,i fllsht paths to diverge. It is necessary, and permissible, for the engint.'er conducting the matching to make slieht adjustments to the input colle,._~tive and fuselage attitude to compensate for minor variations in condili'-~ns. For cxnmplc, provided the limits of collective travel arc r1ot reached. collective n1ay be treated as an internal parameter - it does not limit th~ performarlCC - in practice, the pilot (and the pilot model) will .1djust collc,ct.ive interactively, as required, to sustain some other limiting condition, such as rotor speed or engine torque. Gecause the simulation d0es not atten1pt to model every aspect of the test conditions, the accuracy of the model can only be proved by matching a number of events and checki:-:g that none of the parameters show a consistent error - though a certain amount of scatter is accepted as inevitable. Obviously, the higher the quality of the flight test data, particularly the steadiness of the atmospheric conditions, the easier it is to validate the simulation.

Once the steady state and dynamic accuracy of the vehicle model has been proved, the validity of the pilot model must be confirmed. In the first instance, the time history traces produced during the flight trials are analysed to confirm that the handling technique, in terms of the pitch rates and accelerations used, and the speeds and heights at which events are initiated, etc. , have been correctly defined. The accuracy of the simulation as a performance prediction tool is then proved by attempting to reproduce the flight test results. The program is run with the pilot model "flying" the simulation to replicate the actually flm<n technique, (i.e. using the measured attitudes and rates, if they differ from the prescribed technique) and the model is validated by comparing the distances, drop-downs or whatever is relevant, with the measured results.

The model is intended to predict the performance of a helicopter flown exactly to the laid down technique, in ideal conditions, with a steady wind blowing horizontally at the specified strength throughout the manoeuvre, etc. This state of affairs will never apply in practice. The margins required by the certification authorities (wind factors, flyaway ground clearance, rig miss-distances etc.) are intended to allow for variations from the nominal conditions. When developing techniques, the predicted scatter in performance, due to technique abuse and other factors, must not be bigger than the relevant safety margins. The corollary, is that the margins set by the certification authorities, should be a function of the repeatability of the manoeuvre. I f the margins are significantly larger than are required for safety, the helicopter's payload will be unnecessarily restricted. I t is important therefore, that the handling technique laid down in the flight manual can be repeatably flown by a pilot of average ability, i.e. that small amounts of technique abuse do not have a significant affect on performance. The ease of flying the technique is evaluated by the company and certification test pilots. The repeatability of the technique is one of the factors which is looked for when testing, and when validating the complete model. The ideal is for lhe pr·edictcd performance to lie close to the centre of a small scatter band of flight test results. A significant benefit of using a pilot model is that it is a relatively simple matter to conduct the necessary pare:metric studies to check the consequences of tecllniqtle abuse.

Only when the flight testing and computer validJ.tion lask.s arc complete, and the certification authorities arc s~tisfied that tt1c simulation accurately represents the performance of lhe aircraft, rnr.~y the program be used to run the multitude of cases required Lo create the dataset which will be plotted to produce the dynamic pcrfor·rnanc..e ch<;rts ln

the rotorcraft flight manual. For this work the graphics capability ol· tl1e model is not required and is switched off. The slmulatio11 is scl tiJ) to flJn

as a background task often over night and will calculate tl1c performance for every required combination of aircraft weight, 'dlltude, temperature, wind speed, and obstacle height, etc. A separate suite of computer programs is then used to plot the simulation output data.

4. Structure of t.he Pilot Model

Both the HAPS pilot model and the CHFM HELMSMAN a1·e modular computer programs written in FOHTHAN, with PHIGS graphics sub1·outincs for visualisation. The logic is intended to mimic the th..._>ught process(.:-:s and actions of a human pilot. The pilot model is called at each time step of a simulation run and, using position and rate information from the vehicle model as input, it calculates the control movements required to achieve a specified piloting task. A separate channel of logic iS used for each pilot controllable axis, i.e the pilot model considers the pitch, roll, yaw

TOP LEVEL LOGIC

Pll.OTING LJM:ITATIONS

t

DEFINITION OF REQUIRED MANOEUVRE

't'

VEHICLE UMITATIONS

t

WESTlAND PILOT STRATEGY LOGIC

(DECIDES ON IMMEDIATE PILOTING AJM) CALLS APPROPRIATE HANDLING SUBROUTINE

FOR EACH CHANNEL

VEHICLE RESPONSE

...

-HANDLING SUBROUTINE

--

' " ' PITCH

::-:-!-

R6r.L , YAw·. ''''"~

i:.Q4;E~--

-;:....::.-

-

;::...::_-

-HIERARCHY OF SUBROUTINES WIDCH CALL EACH OTHER BE CAlllNG SUBROUTINE TO

FORE

I GENERATE STICK DISPLACEMENT

INE SUBROUTINES WIDCH DETERM

I REQUIRED ATITIUDE OR VERT! CAL

I ACCELERATION AND THEN CALL I SUBROUTINE TO GENERATE STI CK

D !SPLACEMEI'<'TS

:rE SUBROUTINES WIDCH GENERA STICK DISPLACMENTS DIRECIL y

t:~-i

I - ' ~-~.~ ~ ' ~ ' ~'

1..!--1.

~

..

~ ~

---

-..::-

FEEDBACK---CONTROL lAW I GENERATES FEEDBACK---CONTROL I DISPLACEMENT

s

LOWEST 1 - - -

---

-LEVEL LOGIC

F

~, I

~~

I.

1- I

~~

_~t=

I

f.t=

I ~.~- _{1 VEHICLE}

"=,.;

~--"

"=-:..~~ ~..~~ :....---~ 1~0NSE

:

I+

1..

I+

~

~

-f.

1--

f.

I I ~

,,,,,,It

...

_,

_"'=!:-

I

l:r=l:

~

~

I ~ I

~

_~

-~

I

-- -..=-:

---;:...=. f-_-1 VEHICLE

-crL--~~~~=

F

-

-

·-~-YAW PEDAL LATERAL CYCUC LO:-;GITUDINAL CYCLIC

Figure 4 : HELMSMAN logic flow and calling sequence.

~ ...l

"'

::l 0 ;:;:

"'

...l

g

8

>

and collective inputs independently. The pilot model adapts to changing circumstances and observes any relevant vehicle or piloting 1 irni lations. I f the immediate piloting goal cannot be achieved without exceeding a constraint, the pilot model will amend the manoeuvre in a logical way.

The Westland pilot models simulate the activity of a helicopter pilot at three levels. The top level of the logic can be thought of as modelling the conscious decision making activity of the pilot. Al this

level the pilot knows what the obJect of the exer-cl~~c b>, <UKi Conns a

strategy by which lhe desired end result may be achieved.

The sl.r<\\.egy

b

implemented by setting a series of immediate

pilotin,~

goals, such as

achieving a particular speed and rale of climb. Goal switcl1ing ocCtll's ;1s il manoeuvre develops or in response to unscheduled events. For example the logic will switch as each sub-task is achieved and in response to <1n engine failure or torque limit exceedance. At the second level. the lor;ic models the sub-consclouG activity required to achieve the immeliiatc piJot.in,g goal

set by the lop level logic.

If the Lop level logic sets thco goal of flyin3

in a particular direction, the second level logic will specify \.Jhat. tJu.:~ present roll altitude should be, in order logo from the cui-rent. heading to

lhe target heading.

The lowest levels of the logic can bc thought of as

modelling the instinctive "stick and rudder·" motor skills of the pilot.

These subroutines generate the con t ro 1 d i sp lacernen ts rt.:"qu ired to achieve

the required attitude, torque, rotor speed etc.

Sec figure 4.

What follows is a description of how the pilot m0Jel is implemented

- starting with the lowest level subroutines and working back up to the top

level logic.

The lowest level subroutines are simple feedback control loops which

use error signals to

generate

a control deflection which, when input into

the helicopter model, will result in a vehicle response which tends to

reduce the original error.

The feedback control laws generally consist of

a proportional term for good transient response, and an integral term to eliminate steady-state errors. An error rate term and/or an attitude rate

term is sometimes used to to improve the stability.

The lowest level feedback control algorithms are called by handling

subroutines.

Each handling subroutine has been written to achieve a

specific piloting task; they are the main modular building blocks of the

pilot simulation model.

The handling subroutines combine open loop and

closed loop algorithms, and they are called both by the top level logic and

by each other.

For example, three separate subroutines have been written:

to attain and maintain a specified vertical acceleration, vertical speed,

and height.

They may each be called directly from the top level logic to

generate the collective control inputs required to achieve the relevant

flight condition.

The vertical acceleration subroutine uses a feedback

controller to adjust the collective pitch so as to achieve the required

vertical acceleration. The vertical speed subroutine uses an· open loop

controller to specify what the target vertical acceleration should be, in

the next time step, in order to smoothly attain the required vertical

speed,

The vertical velocity handling subroutine then calls the vertical

acceleration subroutine to generate the required collective control input.

Similarly, the height hold subroutine specifies the vertical speed required

as a function of the height error and then calls the vertical speed

subroutine, which in turn calls the vertical acceleration subroutine, which

generates the control input.

As another example, consider the operation of

the subroutine to attain and hold a specified airspeed.

In this case the

logic adjusts the target pitch attitude in response to the rate of change

of airspeed and then calls one of the lowest level subroutines to generate

the

longitudinal

cyclic

stick

displacement

required

to

match

the

helicopter's attitude with the target value. The maximum pitch attitude, pitch rate, and pitch acceleration values to be used in attaining and

holding the speed are input as data items. The operation of the logic is

illustrated and explained in figure 5.

The lop level of the logic consists of a suite of manoeuvre specific

subroutines.

Each of the top level logic subroutines monitors the progress

of a manoeuvre, observes the vehicle and pilotint.~ limitations, and sets Ulc)

lrnmediate piloting goals. The logic allc.:rnpls Lo achivvc a specified end result, as programmed by the user. The purpo~~e could simply be to lurn

onto a heading at a given rote anci airspeed; or i t could be lo land on a1' offshore platform, on a gusty day, with an engine failur·e al lhc Lmdlng decision point. As with the handling subroutines, the' top level logic subroutines can be 'Written as modules, so lhal very complex manoeuvres can be modelled as a sequence of simpler events. The IIELMSHAN and !lAPS models

SPEED IIOLD LOGIC EXn:RI'-'Al

!)ISHJI\IlA:-.·n::;

I'EIKEI\'ED REOUIJtED PERCEIVED I.ONGJTL!l)INAt.

AIRSI'EF.D AlTITUDE Al'llllJDE <."KLK STK'K

AlRSPEl'.D + s. HA~DLlNG D O

j

FEEDBACK T\r

1

lr

\'E!IJ('!JO

RE!'il'O~SE

HELICOPTER ..._,.__.. MODEL I '

REQUIRED

e

~~[ERROR

1

0;;\C>IC

I'OSIT!t

(SET u~(}-+- SUBROUTING

0-+-1

CONTROL LAW~

TO~~t,T~'JEL 0 L . . . - - - l l'tL01'S 1_;5'r<i~\i1_<£;~ PILOT'S PERCEPT!Ot-1

l

~:;··'~'::·::~~~~~O~F~I~'U~SE~·u~'~G=E==~~~~~:J

1 - AlTITUDE AIRSPE£D PIL01' 1

PERCEHION _MODEL

I

~===""' _{FEEDF.-\Cl\ OF}

PILOTS PERCEPTION OF AIRSPEED m~~~.E

REQUIRED RATE

OF CHA!\'GE

OF AJRS?EED

.ERROR

I

0

AlRSPEED

Required rate of change of rate of change of airspeed (slope) reduced when close to required

airspeed to prevent overshoots

• 1lw Handling Subroutine:

• Uses a feedback controller to determine the required pitch rate as a function of the perceived error in the rate of change of airspeed

• Checks that the proposed pitch acceleration and

pitch rate do not exceed the piloting limits set for the manoeuvre

• Integrates to find the target attitude required for the next time frame

• Checks that the proposed attitude does not exceed the piloting limits set for the manoeuvTe

• Calls a low level handling subroutine to calculate

the stick displacement required to minimise the attitude error

OPERATION OF FEEDBACK CONTROL LAW LOGIC:

e

=

8

_{0 -}

8

( - Error term)

e

= (e- ep)/Llt (-Error rate term)

e

=

e

+

e.t.t ( - Integral term)

1\=

ekt

+

ek,

+

ek

₃

+

(B

_0-

e

)k,

+

ek

₅

0

term only invoked when SAS. is switched OFF ·

improves pitch damping

Figure 5 : Pilot model subroutine logic.

attempt to fly the desired manoeuvre as accurately as possible but, like a human pilot, the pilot model will modify, or even abandon the manoeuvre if vehicle or handling limitations are exceeded. Considr::r the case of an engine failure on take-off - the ''all engines operating'' technique will be flown up to tile moment of engine failure recognition (which could be so1nc

time after the event, to allow for the pilot intcr·vcnlion delay time) - lhc piloting goal will then change, and lhc top level logic >fill either C%Ccutc an OEI continued take-off or landing. Hare subtly, the logl.c will rnc;dify a

manoeuvre lo observe a vehicle 1 tml La lion. In a ~;Lccpl~· b;tnkcd turn, the pilot model will modify lhc manoeuvre t.o ob~;crvc g limit.~; or engine

lorque limits.

In lhis case lhc pilol model will nol t'ly the manoeuvre

exactly o.s specified; the sp<.~ed or helghL may be allm.Jt"'~i to vary in order to obser·ve the more crlllcal limitations. The flight path generated by lhc model will however be reulislic and Hill be clo~>c lo th~1L which a human pilot would t1avc to folloH i11 practice.

HAPS and HELMSMAN read in all of t.he manoeuvre ~'~'''cific simulation

para~:etcrs as data items. When executing a lake-off, f<.~r in~;LlllC't..~. the

airc:--aft weight, the wind speed, llw lake-off decision ~'~)inl (TDI)) height. and :he target speeds etc., can all be varied without makitlg any clliltlncs to the :::omputer code. To generate Lhe dala for the creation of lhe fli[;ht manc.2:.l charts, the simulation is run repeatedly, with each par;unctcr incr-2-mented in turn. All of the relevant input and out~'Ut. parameters arc auto:-.atically recorded for each condition, and are writt~.:~n to a file ready

for ;olotting.

5.

Pilot Model Capability

The CRFM HELMSMAN is still under development but, to illustrate the

current capability of the model

to control helicopter simulations,

a

demo~stration manoeuvre has been programmed - see figure 6. A black and

white representation of the colour interactive display seen by the user, was given in figure 2. For demonstration purposes, the vehicle model used

here is not the complex CRFM, but a

simpler blade model normally used on

DOWNVIIJ','D 800FT

DESCE!\'DL'o'G TIJR.."'' 0\"'TO

. BASE LEG (D

Figure 6 : Demonstration Circuit.

the piloted EH!Ol engineering simulation.

The demonstration begins wi lh

the r.elicopter in the hover above a runway. A take off is performed and the telicopter is flown around a right hand circuit to approach and land on

an elevated helideck using an offshore platform technique.

From lhe hover

above the deck, an offsl1ore platform take-off is performed, with an engine

failt:re recognised just aflcr the TOP lc<1ding to a flyaway.

The pilot

model then takes the aircraft around a left hand circuit, restores the failed engine and flies a normal "all engines operating" approach to lhe runway; returning to the hover at the point where the demonstration bcean. A tir.e history trace for the demonstration circuit is given in fi~urc ~!.

The circuit height, bank angles, pitch and roll rales, c:round tracks, wind speed, etc., are all data items which can be varied. The numbered c:vcnLs on figure 7 refer to the numbered positions marked arotJrld Lt1c circuit sllOW/1

on figure 6.

When used for rolor load predlcllon studies, the CIWfVJIU.M~;MAN will mode 1 manoeuvres of much shor tcr duration than demonstrated above. For example to model limit load cases the pilot model will manoeuvre to achieve a specified fl.lght condition. Performance prediction sludles wlll involve clements of the demonstrallon manoeuvre, such as the t.akc-ofCs a.nd landings. There are other possible applications ho\..'evcr, such ~\S lhe prediction of noise footprints, where elaborate manoeuvr .... ~s, sim\.lar lo the full demonstration, may be required.

"'

8

105]

"

~ ~ :; ~ 98 ~

']

~

~ ~ 0

"'

""]

~

_'

~ 0 0 ~

~

360]

~ 1.0 ~ [:;; iS 0 t 0.5 ~ < : _d ::::

"

"

0

~

']

t: u ~ 100 1:;0

"

~ ~ 0 0 50 u ~ t

1000]

;::

0 t; ;r: 500

d

"

~ 0

SEE FIG. 6 FOR KEY TO NUMBERED EVE/'o.TI

CD G> 0J (4) _{<3J ®} QJ®@ (Ji) 'P (\) (i) QCl Q__(Ji)

20]

0

I

I

I

I I

_I

_I

_{I j}

_I

_{I J.l}

_I

~I ROTO~ S~'Ef:O

I

I I I

'

_'

_L1

I _{I I}

_I

_{ruSE.t.AG£. rrrai:} ATTITUDE (

h.

,-hi

I _{~· 1/}

' ~~

I I

I 'Jt--l

I

' I

_{I I I}

_I

, - I I

_{I I}

_I

i j YA\Y PEDA;L ?OSmON

I

I I

I

i I

I

I

i l l I I I I I I I

I

'

'

I I

I

I I ' ')-_ I I TRV~ AI,RSPEEp I

20]

·20

I

I

I

1'--

I _I

_I

_I

_I

_' I

_{I I l i N}

I

):-, _!\. r-,"\_.!

r...

_I

I

A

l

I

I

! F1JSEU.iJE ROLL AninrDE

I

I

I I

I

1'-i-" I

I

\I l...V

4

_I

_I I

I

I

_'

I II

I _'

I I

I

I

i AIRCRAFT

'j

0.5 0

I

--r--

I

II

II

I I

I

I -:-

H I

I

II~INIG

I I

I

I

I I

_{I I I}

I

I I _FORE & AFT cYcuc Pos'.

1::':

I

I

I

I I

_~ I I I

I I

iAriR.li. CYcuc· ros. 1

'

I I

I

I I

I

I I I I I I I

I

I

I

I

_'

1 I I I ..

l,~t-'

I

I '

_'

_'

_{' I}

1

I

~oLLEmVE

I

'

'

I

I

I

l l I

I

I

I I

_I

~oh--i T<jR01UE~

v

\_I

'

I I I

IL

I I I

I

I ,_:..:=;;.::1

I

I I I

I

_I

I

y

_I

I

t\

..1

rv

E,'<GINE _I\.

v

_{IV' I}

_I

FAD.,URE I I I I

_I

_I

I

v

I

I

I

I

'

I

I

I

I

II

I

Yr

I

I

I I

I

_WHEEL

_HEGIIT

_I

I

'.!/I I I

I

I I

IN-

I

-1 39 79 119 159 199 239 279 319 359 399 439 479 519 559 599 639 679

Figure 7 Computer generated time history trace for demonstration manoeuvre.

6. Conclusions

Westland use helicopter engineering simulations, controlled by pilot

models, for rotor loads and performance prediction studies. The method has several significant advantages.

Pilot models generate realistic flight paths, Hhich can be exactly repeated as many times as necessary; the rnodels are therefore ideal for parametric and technique abuse studies. Pilot model controlled simulations can give the user a very clear insight into wl1at is going on - the engineer can analyse an event in detail, and knows, all of lhe time, exactly what the "pilot" is "thinking" and doing.

Because a human pilot is not included in lhc control loop, it is nol necessary for lhe helicopter simulation to run ln real time - performance models may be run faster lhar1 real ti1nc for ch;1rt d~ta production, and

complex models may be nm al lcs~; \.han real time Cor SU•.:h thl11gs a~~ 1·oto1·

load prediction studies. By maklnn it pos~;ible to run l'(~mplcx mmkls on lO\.J cost workstations, the method makt'S helicopter enginet~ring ~>imul;\Lion

affordable.

At Westland, lhe 2-D HAPS progr;un l£> u~~cd for "~'hiclc design ~'nd development Hork, for predicting Lhe hcl \copter'~; dynamic performance prior to testing, and to pr4

oduce the data for4

fl ir,ht. manual ch~lrts. The

Lhn:-c-dimcnsional CHFM 'Wil.l be used for rotor ;\nd vehicle dcsi~_~n work and, in a simplified form, for dynamic pcrfonnanct~ prcdicLlon stu<iit'S.

Acknowledgements

The HAPS and HELMSMAN programs are the result of many y0~1rs of development

by my predecessors in the Aerodynamics Dc)partmcnt at WHL. I 'Would like Lo

acknowledge the particular contributions of M. llurrhes, \,'ho originated the

HAPS programme, my colleague G. Mat thc\.JS Hho i. s responsible for many recent

enhancements, and D. Swabey whose skill with PHIGS is self evident.

I am

also grateful to F.J. Perry and R.E. Hansford for valuable suggestions and

to

Westland's

test

pilots,

in

particular

J.

Tracy,

for

thoughtful

explanations and practical demonstratiOllS of piloting tecl1niques.

The development of the Helmsman module has been funded by the United

Kingdom Ministry of Defence (Procurement Executive), under the contractual

supervision of DRA (MS2).

References

1.

R. E. Hansford, The development of the coupled

rotor-fuselage model

(CRFM), American Helicopter Society 48th Annual Forum, June 1992.

2.

P. T. W.

Juggins, A comprehensive approach to coupled rotor-fuselage

dynamics, Forum Proceedings of the 14th European Rotorcraft Forum,

Paper No.48, September 1988.

3.

P.T.W.

Juggins,

Application

of

the

WHL

CRFD

program

to

total

helicopter dynamics, American Helicopter Society National Specialists'

Meeting on Rotorcraft Dynamics, Arlington, Texas, November 1989.

P.T.W

Juggins,

Coupled

understanding Helicopter

Annual Forum, June 1992.

rotor-fuselage mode

shapes

A tool

in

response, American Helicopter Society 48th